The ubiquitin ligase FBXW7 targets the centriolar assembly protein HsSAS-6 for degradation and thereby regulates centriole duplication

Formation of a single new centriole from a pre-existing centriole is strictly controlled to maintain correct centrosome number and spindle polarity in cells. However, the mechanisms that govern this process are incompletely understood. Here, using several human cell lines, immunofluorescence and structured illumination microscopy methods, and ubiquitination assays, we show that the E3 ubiquitin ligase F-box and WD repeat domain–containing 7 (FBXW7), a subunit of the SCF ubiquitin ligase, down-regulates spindle assembly 6 homolog (HsSAS-6), a key protein required for procentriole cartwheel assembly, and thereby regulates centriole duplication. We found that FBXW7 abrogation stabilizes HsSAS-6 and increases its recruitment to the mother centriole at multiple sites, leading to supernumerary centrioles. Ultrastructural analyses revealed that FBXW7 is broadly localized on the mother centriole and that its presence is reduced at the site where the HsSAS-6–containing procentriole is formed. This observation suggested that FBXW7 restricts procentriole assembly to a specific site to generate a single new centriole. In contrast, during HsSAS-6 overexpression, FBXW7 strongly associated with HsSAS-6 at the centriole. We also found that SCFFBXW7 interacts with HsSAS-6 and targets it for ubiquitin-mediated degradation. Further, we identified putative phosphodegron sites in HsSAS-6, whose substitutions rendered it insensitive to FBXW7-mediated degradation and control of centriole number. In summary, SCFFBXW7 targets HsSAS-6 for degradation and thereby controls centriole biogenesis by restraining HsSAS-6 recruitment to the mother centriole, a molecular mechanism that controls supernumerary centrioles/centrosomes and the maintenance of bipolar spindles.

Maintenance of correct centrosome/centriole number in cells is crucial for genetic stability and tissue homeostasis (1,2). In the proliferating cells, each centriole of an interphase centrosome duplicates strictly once during the cell cycle to gener-ate two centrosomes that direct bipolar spindle assembly, a prerequisite for faithful segregation of chromosomes (3,4). Failure of centriole duplication leads to mitotic cells with monopolar spindles, whereas its overduplication results in supernumerary centrosomes and multipolar spindles (5)(6)(7). During G 1 /S entry of the cell cycle, each mother centriole starts assembling only a single new centriole, called the procentriole, at a specific site toward its proximal end. The molecular processes that control formation of only one procentriole per mother centriole remain poorly understood.
Recent studies have implicated roles of ubiquitin-mediated protein degradation in regulating the functions of centrosome/ centrioles (28,(32)(33)(34)(35). One such key regulator is SCF FBXW7 with its substrate targeting subunit FBXW7, an evolutionarily conserved E3 ligase (36). FBXW7 is mutated in numerous cancers and is linked to chromosomal instability and tumorigenesis (36 -38). Recently, FBXW7 abrogation has been shown to negatively regulate cilia assembly, a process that requires functional integrity of the centriole (39). Furthermore, FBXW7 expression bears a negative correlation with centriole number in cells (40). However, the molecular link of FBXW7 to centriole amplification is not clearly understood.
Here, we demonstrate that SCF FBXW7 targets HsSAS-6 for degradation and thereby controls centriole biogenesis by restraining HsSAS-6 recruitment to the mother centriole.
Ultrastructural analyses reveal direct association of FBXW7 with HsSAS-6 and its unique localization on mother centriole that ensures HsSAS-6 recruitment to a specific site to form the new centriole. Biochemical results demonstrate ubiquitin-mediated degradation of HsSAS-6 by SCF FBXW7 . Our data for the first time provide a molecular basis for the control of supernumerary centrioles and centrosome amplification in human cells.

FBXW7 controls centriole duplication by regulating HsSAS-6 at the centrioles
We first analyzed centriole numbers both in FBXW7 knockout (KO) FBXW7 Ϫ/Ϫ DLD-1 cells (38) and in HeLa cells depleted of FBXW7 by esiRNA. Formation of extra (more than four) centrioles was induced in both cases in the G 1 /S synchronized condition ( Fig. 1 (a and b) and Fig. S1 (a and b)). The percentage of cells with more than four centrioles was ϳ3-fold higher (19% versus 6%) in FBXW7 KO cells as compared with FBXW7 WT cells ( Fig. 1b and Fig. S1b). We then inspected how loss of FBXW7 affects procentriole assembly proteins at the centrosome. HsSAS-6 was found to be localized specifically to each of the supernumerary centrioles in the KO or depleted cells ( Fig. 1c and Fig. S1 (c and d)). Total centrosomal HsSAS-6 intensity was increased (Fig. 1d). Plk4 appeared to localize to each of those centrioles, but its overall level did not seem to increase (Fig. S1e). Co-staining of HsSAS-6 with mother centriole marker, Odf2, showed multiple HsSAS-6 foci surrounding a single Odf2-stained centriole, indicating that the extra procentrioles were generated on a single mother centriole (Fig.  1e). In concurrence with supernumerary centrioles, formation of extra centrosomes and multipolar spindles were induced (Fig. 1, f and g).
We next assessed the structural organization of FBXW7 at the centrioles. Structured illumination microscopy (SIM) 2 imaging of centrin-1-eGFP stable RPE-1 cells revealed that FBXW7 localizes to both the centrioles radially by covering most of the areas, but less in the areas where HsSAS-6 is localized (Fig. 3a). Three-dimensional reconstructed images showed a clearer view of such organization (Fig. 3c). A model representing FBXW7 and HsSAS-6 localization is shown in Fig.  3b. The absence of FBXW7 at and near the procentriole sites was further confirmed by co-staining FBXW7 with centrobin, a daughter centriole marker (41) (Fig. 3d). However, as the centrioles were oriented randomly, it was difficult to infer from the SIM images whether FBXW7 localizes along the whole length of the centriole or in the region at and near its proximal end. Because in normal cells, FBXW7 localization was somewhat inversely correlated with HsSAS-6, we then checked whether the two proteins could associate with each other in the HsSAS-6-over-expressed condition. We imaged FBXW7 in eGFP-HsSAS-6 -overexpressed HeLa Kyoto cells under treatment of protease inhibitor MG-132. Endogenous FBXW7 strongly colocalized with eGFP-HsSAS-6 at and around the centrioles (Fig. 3e). FBXW7 localized around the eGFP-HsSAS-6stained centrioles as multiple clusters. Overall, these findings indicate that FBXW7 directly associates with HsSAS-6 at the centriole. We also imaged HsSAS-6 localization on the mother centriole in the FBXW7 KO DLD1 cells. Unlike the FBXW7 WT cells, the knockout cells showed HsSAS-6 localization at multiple sites on and around the mother centriole ( Fig. 3f), suggesting deregulation of Hs-SAS-6 recruitment on the mother centriole.

FBXW7 targets HsSAS-6 for degradation
We next assessed whether FBXW7 regulates HsSAS-6 stability. The level of HsSAS-6 was increased significantly in FBXW7-depleted HeLa and U2OS cells (Fig. 4, a and b) and similarly in FBXW7 knockout DLD-1 cells (Fig. 4c). A ϳ1.5-1.8-fold increase of HsSAS-6 level as compared with control was observed in all of these cell lines (Fig. 4c). However, FBXW7 depletion did not alter Plk4 or FBXW5 (Fig. S2d) (32). We found that HsSAS-6 stabilization was pronounced maximally at G 1 /S followed by S phase (Fig. 4d). A rescue experiment by expression of exogenous FLAG-FBXW7 in the knockout cells reduced HsSAS-6 to a level comparable with that of FBXW7 WT cells (Fig. 4e). We next determined the role of the WD40 substrate-binding region (Fig. S1f). Exogenously expressed Myc-HsSAS-6 level was more stabilized in FLAG-FBXW7-WD⌬ cells as compared with FLAG-FBXW7 WT (Fig. 4f). Plk4 and FBXW5 did not change under similar conditions (Fig. S2e).

FBXW7 physically interacts with HsSAS-6
We next determined whether FBXW7 interacts with HsSAS-6. Co-immunoprecipitation (co-IP) in G 1 /S synchronized HeLa and HEK293T cells showed the presence of HsSAS-6 in the immunoprecipitate of endogenous and FLAGtagged FBXW7, respectively (Fig. 4, g and h). The interaction was further confirmed by reverse co-IP of endogenous HsSAS-6 in HEK293T cells (Fig. 4i). Furthermore, FLAG pulldown of WD domain-deleted FBXW7 showed no HsSAS-6,

ACCELERATED COMMUNICATION: Centriole duplication by FBXW7
indicating that the WD domain is required for the interaction (Fig. 4h).

FBXW7 ubiquitylates HsSAS-6
We next determined whether SCF FBXW7 ubiquitylates HsSAS-6. We first assayed ubiquitination in vivo in HEK293 cells (see "Experimental procedures"). Levels of ubiquitylated proteins were substantially increased in the FLAG-FBXW7expressed cells (Fig. 5a). Unlike FBXW7 WT, FLAG-FBXW7-WD⌬ cells showed a very low level of ubiquitination, comparable with the control cells (Fig. 5a). We also checked the ubiquitination of purified recombinant MBP-tagged HsSAS-6. All of the four SCF FBXW7 complex components were purified from HEK293T cells by FLAG pulldown and mixed with MBP-HsSAS-6 ( Fig. S2f) (42). MBP-HsSAS-6 was ubiquitinated in a dose-dependent manner (Fig. 5b). Ubiquitin-mediated destruction of substrate proteins involves formation of polyubiquitin chains through covalent conjugation through Lys residues at 48 and 63 (43). We tested involvement of those two lysines of ubiquitin in FBXW7-mediated HsSAS-6 ubiquitination. A mixture of purified MBP-HsSAS-6 and SCF FBXW7 complex was incubated with either Ub WT or Ub K63R or K48R mutant in the presence of E1, E2, and ATP. HsSAS-6 ubiquitination was substantially reduced both in the Ub K63R and Ub K48R mutants as compared with Ub WT (Fig. 5c). Altogether, these results demonstrate that HsSAS-6 is a ubiquitylation substrate of SCF FBXW7 .

Putative phosphodegron sites in HsSAS-6 are involved in FBXW7 interaction
FBXW7 targets substrate proteins through the substrate consensus Cdc4 phosphodegron (CPD) motif with the sequence (S/T)PX 2 (S/T/D/E/X) (where X represents any amino acid) (36,44). In many cases, prior phosphorylation in the CPD sites is involved in substrate recognition by the ligase. Sequence analysis revealed two conserved CPD-like motifs in HsSAS-6, one at Ser-111 (SPAAI) and the other at Thr-495 (TPPAH) (Fig. S2g). We mutated these putative CPD sites Ser and Thr to alanine (2A) and checked FBXW7-HsSAS-6 interaction and its ubiquitination. FLAG-FBXW7 interaction with Myc-HsSAS-6 was markedly reduced in the case of 2A mutant as compared with WT Myc-HsSAS-6 (Fig. 5d). An
The number of centrioles/centrosomes is tightly controlled to ensure bipolar spindle assembly and genome integrity. Here, we have identified a hitherto unknown molecular mechanism that governs this process. Ultrastructural analysis of cells with overexpressed HsSAS-6 evidenced formation of multiple procentrioles on the mother centriole (28), indicating that nucleation of procentriole, in principle, can occur at multiple sites on the mother centriole. SAS-6, being the major constituent of the procentriole cartwheel, is a key player for activating the process. Our data demonstrate that FBXW7 regulates HsSAS-6 level, both at the overall cellular level and at the centrosome (Figs. 2 and 3). This is likely to be mediated by ubiquitination-mediated degradation, as our data demonstrate that HsSAS-6 is ubiquitinated by SCF FBXW7 (Fig. 5). Concurrent with these findings, our SIM data have supported a direct centriole-specific targeting of HsSAS-6 by FBXW7. In HsSAS-6 -overexpressed cells, FBXW7 shows direct association with the centriole-localized HsSAS-6. Interestingly, the ligase appeared as multiple clusters around HsSAS-6 and centriole. Such clustering may indicate localization of FBXW7 in the form of large structures in the centriole for targeted destruction of HsSAS-6. Contrastingly, at normal conditions, however, FBXW7 showed an inverse localization pattern with respect to HsSAS-6. One explanation could be that most of the endogenous HsSAS-6 was degraded by the ligase before it could even accumulate to the mother centriole. The only exception is the procentriole assembly site, where HsSAS-6 recruitment is facilitated, but FBXW7 localization is minimal. However, determination of how FBXW7 localization is inhibited at the procentriole site requires future investigation. Overall, our data indicate that FBXW7 plays a major role in defining the site of procentriole assembly on the mother centriole.
HsSAS-6 protein level is controlled by degradation by APC/ C-Cdh1 during late mitosis until G 1 (28) and by another F-box family ligase, SCF FBXW5 , during late S phase (32). However, APC/C-Cdh1 is inactivated as the cells progress to G 1 /S due to degradation of its cyclin counterpart (45,46). Similarly, SCF FBXW5 activity is diminished sharply after G 2 due to FBXW5 degradation by APC/C-Cdc20 (32). Moreover, activity of Plk4 is high during G 1 /S (33,47,48). The FBXW7 level is high at G 1 until G 1 /S and starts to decrease as the cells proceed through S phase (40). We find here that HsSAS-6 degradation is maximal at G 1 /S until early S phase (Fig. 3). This indicates that FBXW7-mediated centriole duplication control is primarily active during G 1 /S until early S phase.
In conclusion, we have demonstrated that targeted degradation of HsSAS-6 by SCF FBXW7 is a key mechanism that controls centriole amplification.

Co-IP
Cell lysates were incubated with antibodies as specified for immunoprecipitation, and proteins were pulled down.

Immunofluorescence microscopy
Methanol-fixed cells were incubated with primary antibodies followed by incubation with secondary antibodies. The images were captured by a Leica SP 5 laser confocal microscope.

SIM
The images were captured through a ϫ100 1.49 numerical aperture Apo Oil objective in an N-SIM microscope (Nikon).

ACCELERATED COMMUNICATION: Centriole duplication by FBXW7
More details of the experimental procedures are provided in the supporting Experimental procedures.